Enhancing Breast Cancer Classification through Attention Based VGG-19 and Federated Learning with Multi-Center Medical Imaging

Breast cancer remains the primary cause of death for women worldwide, which emphasises how important it is to detect the disease early and accurately in order to increase survival rates. Conventional approaches encounter substantial challenges since large labelled datasets are required, and they are frequently limited by privacy issues and dispersed data sources. This work introduces an novel method to improve breast cancer classification by combining Attention-based VGG-19 with Federated Learning in multi-center medical imaging. Accurate breast cancer detection relies on analyzing medical images such as mammograms, MRIs, and biopsies, which often vary across different healthcare centers. The proposed model uses VGG-19, a deep convolutional neural network, enhanced with an attention mechanism to highlight key regions in the images for more precise classification. Federated Learning is also integrated, enabling multiple medical institutions to collaboratively train the model without sharing sensitive patient data, thereby preserving privacy and data security. This approach allows the model to learn from diverse datasets while keeping the data localized. The synergy of attention-based VGG-19 and Federated Learning enhances the accuracy, robustness, and generalization of breast cancer classification, offering a promising solution for early detection and diagnosis across multiple medical centres.